Peak Shape Calibration Method Improves the Mass Accuracy of Mass Spectrometers

ABSTRACT

The mass accuracy of unit resolution mass spectrometers (MS) is generally near 0.2 Da (~400 ppm at 500 Da) and reported to the nearest integer mass. This is because the software reduces the profile mode spectrum to a centroid spectrum with great loss of information. In this article we will introduce a novel way to calibrate this class of MS instruments that allows routinely achieving 5–10 ppm mass accuracy as well as dramatically improving ion signal extraction and better ratios of signal to noise (S/N).

Centroid Mode Calibration

Three parameters express the mass (σ_ppm) for any mass spectrometer (MS) instrument. It is inversely related to the resolving power (R) of the instrument, the square root of the signal strength (S), and a constant (C) , as shown in this equation,^1,2

Equation 1

The constant is determined by factors such as signal unit conversion to real ion counts, peak sampling interval, peak analysis, and mass determination algorithms. The mass accuracy of unit resolution instruments should approach that of high-resolution instruments, but it does not. Instruments based on quadrupole or ion trap technology typically have R values of ~1,000 whereas a high-resolution instrument such as a hybrid quadrupole Time of Flight (qTOFMS) unit has an R value of ~5,000. In theory, the high ion transmitting efficiency (S) of the lower resolution unit should narrow the gap in resolution. In practice, however, the low unit resolution instruments demonstrate a mass accuracy level between 1 and 2 orders of magnitude worse, pointing to an unusually high C value on low-resolution systems.

To understand what impacts the constant, it is first necessary to understand the current approach typically used in acquiring MS data. In most cases, data presented to the user at the end of the acquisition is centroid data, the well-known "stick" graph, in which an ion's m/z (molecular mass/charge) positions are represented by single lines on the graph (Figure 1). This form of graph represents the MS data after heavy processing by the instrument firmware or post-analysis software.

Figure 1 is a flow diagram of a typical instrument data acquisition process. The detector signal is sampled across time (or pixel) to obtain the raw continuum data as output from the detector. A previously established linear or nonlinear calibration equation within the instrument firmware transforms the X-axis to m/z (mass only calibration). This results in what is commonly referred to as a profile mode spectrum.

While most instruments can collect data in the profile mode, it is more common when using unit resolution instruments to let the instrument firmware attempt to locate the peaks from the profile data, and then produce the centroid or stick spectrum before presenting the data to the analyst. This makes for easy data transmission, storage, and management due to the high data rate in a MS system. It made sense in the days of older computer technologies.

There are a number of flaws in this approach. One of the biggest flaws is that peak positions cannot be located accurately. If the peak positions are not accurately and reproducibly determined, this inaccuracy can lead to substantial error both in creating the calibration and in applying the calibration. If one examines a typical profile mode spectrum (Figure 2), it is easy to see why there can be substantial errors in peak locations. First, there is substantial asymmetry in the peaks. This asymmetry is related to the instrument "tune" and typically varies across the spectrum and from one instrument to another. This makes determination of the peak position prone to substantial error, especially at low resolutions. Secondly, there is the presence of noise, which can also lead to substantial errors.

Unfortunately, since the default mode is centroid-mode data acquisition, the data presented to the analyst have been processed through various error-prone centroding algorithms. This results in significant and (most of the time) unrecoverable information loss. It is therefore no surprise that the mass accuracy specified for conventional unit mass resolution MS systems is typically no better than 0.1 or 0.2 Da.

A NEW APPROACH: CALIBRATING PEAK SHAPE

Based on these limitations, we concluded that calibration only on the X-axis would be insufficient to obtain the best mass accuracy. Instead, a novel calibration approach was developed that not only calibrates the X-axis, but also calibrates the peak shape. This is possible in MS because the theoretical peak positions are known very accurately (they are simply the mass, or more precisely, m/z of each ion). In addition, we also know quite accurately the various isotopes and their relative abundances. This information allows the calculation of a calibration function that corrects not only the peak position, but also the peak shape.

We named this novel calibration technique "MSIntegrity," which calibrates the instrument to a symmetrical peak shape. This method can dramatically improve all downstream processing. For example, we can perform accurate peak picking (the algorithmic process of locating the center and integrating the area of a spectral peak) with no user-set parameters due to the calibrated peak shape, which allows accurate peak center and peak area to be determined. As a side benefit, we found that noise can be substantially suppressed without the loss of information and peak distortion typical of commonly used smoothing operations.

Figure 3 illustrates the process of applying the MSIntegrity calibration. Although the process flow is identical to the conventional approach, there are significant distinctions that preserve the rich information contained in a MS scan. A comprehensive calibration involving both the mass and peak shape is ideally performed through on-board instrument processing to produce fully calibrated MS data in profile mode. All downstream analysis, including accurate mass determination, will be based on this fully calibrated MS data in profile mode. Alternatively, the calibration can be done after collecting profile mode data in the MassWorks software implementation of MSIntegrity.

POTENTIAL APPLICATIONS

One application of this calibration technique is shown in Figure 4 for liquid chromatography- mass spectrometry (LC/MS). In this example a typical unit mass resolution triple quad instrument is used to screen for the metabolites of the drug verapamil after microsome incubation. Since the calibration was performed after the data acquisition, the known compound (verapamil) could be used as the calibration ion for the entire run. The chromatographic peak eluting before verapamil is suspected of being its demethylation metabolite. To confirm this, the accurate mass of the calibrated monoisotope peak at 441.2742 Da compares favorably to the theoretical mass of 441.2753 Da. This corresponds to an error of about 1 mDa or 2.5 ppm. More typically, we have found that we can routinely measure the mass accuracy on such instruments in the 5-10 ppm range.

Figure 5 shows an example of the calibration applied to an environmental gas chromatography – mass spectrometry (GC/MS) application for identifying pesticides using an Agilent MSD detector (single quad mass spectrometer). A commercial calibration mixture was used to externally calibrate the MS. In the blind test, the Monsanto pesticide PCB 209 (CAS number 2051-24-3) was calculated as having a mass of 493.6855 Da as compared to the theoretical mass of 493.6885 Da. This corresponds to an error of 3 mDa or about 6 ppm, a mass accuracy observed for other pesticides in the mixture as well.

Another application area includes the analysis of a known compound in spectra containing a complex background. For example, in metabolite identification, drug metabolites need to be identified in extremely complex bile matrices. Conventional extracted ion chromatograms (XIC) are not selective enough, and background interferences create many false positives. Each positive may need to be further investigated with techniques such as MS/MS or C14-labeled compounds.

Due to the comprehensive calibration of the MSIntegrity systemn, we can instead calculate an XIC by using a narrower and more accurate mass window, and the entire isotope profile can be used as a highly selective mass filter. The accurate mass profile XIC (AMPXIC) provides a dramatic improvement by rejecting interferences including many that actually fall within the same mass window. An example is shown in Figure 6 with verapamil incubation. This same approach can be applied to impurity, degradant, and compound identification studies.

CONCLUSION

The ability to obtain high mass accuracy on unit mass resolution instruments combined with peak shape calibration has the potential to enhance or even create entirely new methods of analysis. One benefit of this approach is that it allows the input of calibrating-data from various instruments (including instruments of various types made by different manufacturers) to output the same results in terms of peak shapes and positions regardless of instrument tune and other variables (mass spectral alignment). Thus, applications that are heavily dependent on identifying subtle differences between complex samples, such as biomarker discovery and proteomics, can be dramatically improved.

The MSIntegrity approach to MS calibration can dramatically extend the utility of both new and existing unit mass resolution systems by allowing high mass accuracy measurements to be performed for compound identification or verification. In addition, the calibration improves the form and consistency of the data from any mass spectrometer.

Quick Recap

ACKNOWLEDGEMENTS

The authors wish to thank Xian-guo Zhao and Zhe-ming Gu from XenoBiotic Laboratories, Inc., for providing the metabolite LC/MS data.

Don Kuehl, PhD, is vice president, marketing and development, at Cerno Bioscience, 14 Commerce Dr., Danbury CT 06180, tel. 203.312.1150, fax 203.312.1159, info@cernobioscience.comYongdong Wang, PhD, is founder and president of Cerno Bioscience.

REFERENCES

1. Blom KF. Estimating the Precision of Exact Mass Measurements on an Orthogonal Time-of-Flight Mass Spectrometer Anal. Chem. 2001;73:715-719.

2. Moini M, et al. Sodium trifluoroacetate as a tune/calibration compound for positive-and negative-ion electrospray ionization mass spectrometry in the mass range of 100-4000 Da. JASMS; 1998( 9):977.

3. Gu M. Wang Y, Zhao X, Gu Z. Accurate mass filtering of ion chromatograms for metabolite identification using a unit mass resolution liquid chromatography/mass spectrometry system. Rapid Comm. Mass. Spec. 2006; 20(5):764. Published online 2006 Feb 6.